Article Preview
TopIntroduction
During the Great East Japan Earthquake, communication systems in affected areas were damaged catastrophically. Moreover, telecommunication carriers emplaced communication restrictions such as bandwidth limitations to avoid traffic congestion. Under these circumstances, it was extremely difficult to communicate smoothly among users, including the direct victims of the disaster. Especially, information exchange necessary to confirm the victim’s safety, i.e., safety confirmation information, which was the most important communication during the disaster, was difficult. Many people were put in an uneasy situation for a long period. Safety confirmation in disasters can be done with short message exchanging and sharing. Therefore, it can be served by the remaining terminals and networks. Nevertheless, numerous important messages were sent through an unstable network environment that had been partially destroyed. Service should have been provided with a low network load.
To realize stable network services in such cases, P2P networks have been attracting attention as a communication infrastructure. P2P networks are application-layer networks in which each node manages contents in a distributed manner. Because these networks decentralize communications to each node, they have high scalability and robustness. These characteristics are expected to fit the requirements for use with unstable network environments in such disasters.
P2P networks are classifiable into unstructured P2P networks and structured P2P networks. It is easy for nodes in structured P2P networks to communicate each other effectively. Moreover, contents in nodes can be searched reliably. Particularly, many algorithms of re-construction of networks have been proposed to realize dynamic load balancing according to network environment changes and node performance. However, in existing algorithms of load balancing for structured P2P networks, some restrictions must be set in network structures. This limitation makes it difficult to achieve functional diversity.
In contrast, we have proposed a structured P2P network, the Well-distribution Algorithm for an Overlay Network (WAON) (Takeda et al., 2012). WAON provides a mode of effective information exchange by building a cyber space, deeply considering the situation of a real space. This approach is based on the concept of Symbiotic Computing, which aims at the integration of real space and cyber space (Suganuma, Miyamoto, Makishi, Kitagata, & Shiratori, 2007; Suganuma, Sugawara, & Shiratori, 2007; Suganuma, Sugawara, Kinoshita, Hattori, & Shiratori, 2009a; Suganuma, Takahashi, Izumi, Kinoshita, & Shiratori, 2009b) in the context of cognitive informatics. WAON achieves dynamic load balancing according to changes of the network environment and performance differences of the nodes, with no network structure limitations. In this method, inter-node communication costs can be reduced because each node can perform load balancing without network reconstruction. In addition, WAON can reflect the physical network topology in the real space; thereby it can reduce physical network traffic. These features of WAON enable practical applications to work in structured P2P networks with unstable network environments.
As described in this paper, we propose a P2P safety confirmation system based on the WAON. The proposed system is a file-sharing system that is assumed to be used in times of large-scale natural disasters. This system constructs an application-layer overlay network using remaining network equipment and terminals, and exchanges safety confirmation information related to that. The safety confirmation information includes personal profiles of disaster victims such as name, age, evacuation site, etc. Users can search that information. The information is forwarded and stored in a distributed manner in several nodes in the network, considering load balancing among nodes.
Features of the WAON-based P2P safety confirmation system are as follows: